Voice-off: Siri vs. Android

Android had voice commands first, but Apple's Siri comes on strong. Can Android 4 and Windows Phone 7 'Mango' keep up?

The InfoWorld voice-recognition challenge

The iPhone 4S's Siri has captured the public's imagination, yet Android has long supported voice-based search, as does Windows Phone "Mango." So which is best? InfoWorld had Ken Marquis -- a former actor, TV producer, and frequent emcee with that "radio voice" -- test several queries to give each device its best chance to interpret and respond correctly.

We soon discovered "Mango" can't take dictation except for text messages, so our dictation voice-off is between just Siri and Android. But "Mango" competed in our response voice-off.

Related reviews:
Apple iPhone 4S
Samsung Galaxy Nexus
Samsung Focus S
Comparison: Smartphone face-off

Dictation test 1: Picking peppers (Siri)

Even for Marquis, this was a tongue twister:

Peter Piper picked a peck of pickled peppers,

A peck of pickled peppers Peter Piper picked;

If Peter Piper picked a peck of pickled peppers,

Where's the peck of pickled peppers Peter Piper picked?

Still, Siri got everything right.

Dictation test 1: Picking peppers (Android)

As in every test, Marquis spoke this tongue twister with all devices ready to listen, rather than say the phrase separately for each device:

Peter Piper picked a peck of pickled peppers,

A peck of pickled peppers Peter Piper picked;

If Peter Piper picked a peck of pickled peppers,

Where's the peck of pickled peppers Peter Piper picked?

Android got one word wrong (highlighted at left), and it didn't know Peter is a proper name.

Dictation test 2: Shopping list (Siri)

This list includes common words, brand names, and foreign names to test the vocabulary range:

Grocery list: bananas, milk, cereal, sandwich bread, pickles, Camembert, tofu, eggs, leg of lamb, hummus, Wheaties, Craisins, chocolate milk, onions, fennel, daikon radish, jicama

Siri missed "daikon" and "jicama."

Dictation test 2: Shopping list (Android)

Again, this list includes common words, brand names, and foreign names to test the vocabulary range:

Grocery list: bananas, milk, cereal, sandwich bread, pickles, Camembert, tofu, eggs, leg of lamb, hummus, Wheaties, Craisins, chocolate milk, onions, fennel, daikon radish, jicama

Android missed "Camembert" and "jicama," and didn't know "Wheaties" and "Craisins" are proper names.

Dictation test 3: Gettysburg address (Siri)

This tests real dictation for note-taking and memo-writing:

Four score and seven years ago our fathers brought forth on this continent, a new nation, conceived in Liberty, and dedicated to the proposition that all men are created equal. Now we are engaged in a great civil war, testing whether that nation, or any nation so conceived and so dedicated, can long endure. We are met on a great battlefield of that war. … (Read Abraham Lincoln's full Gettysburg address.)

Siri made three mistakes; the highlighted words at left should be "endure," "shall," and "earth."

Dictation test 3: Gettysburg address (Android)

In its transcription of the Gettysburg address, Android made more than a dozen errors, some shown at left.

Response test 1: Appointments

Query phrase: When is my next meeting?

Siri checked the calendar for the next appointment, and it both spoke and displayed the results.

Android 4 and Windows Phone "Mango" both did a Web search. Android searched the full phrase (and displayed mainly results on how to talk to Siri, ironically), whereas "Mango" searched just for "meeting," with random results.

Response test 2: Directions

Query phrase: Directions to 501 Second St.

Siri responded by saying, "Here are the directions to 501 Second Street," and showed directions on a map to that address in the current city (San Francisco), though it may provide a list of multiple nearby cities in some cases.

Android could find no matches and displayed no map when asked in the Android browser, though when asked in the Navigation app, it displayed several addresses in nearby cities to choose from, then showed the directions.

"Mango" didn't understand the request the first time. On the second try, it displayed the address (in the current city) on a map, but no directions.

Response test 3: Hotel reservation

Query phrase: Search for hotels in Carmel with an oceanfront view available in February or March.

Siri didn't understand the query the first time. On the second try, it displayed a list of hotels in Carmel, Ind. -- where there is no oceanfront. It did not check availability. When we added "California" to the query, it displayed hotels in Carmel, Calif., but didn't filter by availability or view. From that point on, it assumed California when we said just "Carmel."

Android displayed a list of hotels in Carmel, Calif. "Mango" didn’t understand the first try but responded with a list in the second attempt, also for Carmel, Calif.

Response test 4: Weather

Query phrase: How cold is it outside?

Siri spoke the predicted low for the day in the current location -- we were expecting the current temperature -- and showed the seven-day weather forecast on screen. When asked, "What's the current temperature outside?" Siri spoke the current temperature and showed the weather forecast. When asked, "How hot will it be today?" Siri spoke the expected high temperature and showed the forecast. It's clear that Siri's range of response to questions can be more nuanced than you might expect.

Android searched the Web for "current temperature," and "Mango" searched for "outside."

Voice services compared

In Siri, long-tap the Home button or raise the iPhone 4S to your ear. When you hear the dink sound, state your query or command. Siri interprets it, then displays a transcription, results (also spoken), and info from relevant apps. For active text fields, tap the microphone icon to transcribe whatever you say.

In Android, tap the microphone icon in the search field to initiate voice-based search. Google opens and shows the search results. In the Navigation app, tap the Speak Destination button to tell the app what address you want directions to. In active text fields, tap the microphone icon to begin transcription.

In "Mango," long-tap the Start button to initiate a voice-based search or a command to send a text message.